NewsBite analysis · AI, Business
VB
85% of companies burned by an AI mistake are racing to cut the humans who might catch the next one
A recent survey indicates that enterprises which have previously encountered issues with AI deployments are hastening the removal of human involvement in decision-making processes. This shift occurs even as trust in automated evaluations increases, suggesting a paradox where organizations are leaning towards automation to mitigate past failures. Despite improved confidence metrics, significant risks remain as many companies proceed with deployment practices that may overlook critical checks for AI correctness.
MonitorConfidence: MediumShort term horizonTalent displacement risk: High
Published 18 Aug 2026, 16:25 UTC10 min read
Visit original article ↗Lead image — VB
Summary
TL;DR
- 85% of companies affected by AI issues are removing human oversight.
- Trust in automated evaluations rose from 5% to 13% in one month.
- 49% of organizations reported AI features passing tests that later failed with customers.
- The reliance on automated checks is increasing despite past failures.
- A significant portion of companies continue to allow AI to make changes without human consent.
Interesting facts
- Confidence among firms with no reported failures contrasts sharply with those that have experienced setbacks.
- The survey suggests a growing trend of firms adopting more monitoring solutions instead of improving evaluation sets.